Remote sensing image target detection method based on attention mechanism and lightweight convolution

By introducing Biformer's attention mechanism and lightweight convolution GSCONV in the remote sensing image target detection, combined with Slim-Neck and SPPCSPC structures, improved the YOLOV8 network model, solved the low accuracy and recall rate of traditional algorithms in small targets and multi-scale target detection rates. The problem achieves higher detection accuracy and robustness.

CN118781465BActive Publication Date: 2025-01-28LIANGJIANXINGTONG (SHENZHEN) TECHNOLOGY CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202410808919.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-06-21
Publication Date
2025-01-28
Estimated Expiration
2044-06-21

AI Technical Summary

Technical Problem

Traditional remote sensing image object detection algorithms have problems with low detection accuracy and recall when dealing with small targets and multi-scale targets, which are prone to missed detection and missed detection.

Method used

Adopt the Biformer's BiFormer's attention mechanism and lightweight convolution GSCONV, combined with the SLIM-Neck design paradigm and the SPPCSPC space pyramid structure to improve the YOLOV8 network model to improve the accuracy and robustness of the target detection.

Benefits of technology

It has significantly improved the detection accuracy of the target detection model, enabling it to more accurately locate and identify the target object, and improve the robustness and real -time detection capabilities of the model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118781465B_ABST
    Figure CN118781465B_ABST
Patent Text Reader

Abstract

The present invention provides a remote sensing image target detection method based on an attention mechanism and lightweight convolution, wherein the method comprises: step 1: obtaining an optical remote sensing image target detection data set; step 2: embedding an attention mechanism in a YOLOv8 backbone network; step 3: replacing the standard convolution of the YOLOv8 backbone network with lightweight convolution, and introducing the Slim‑Neck design paradigm; step 4: fusing the SPPCSPC structure into the feature extraction layer in the YOLOv8 framework to obtain an improved YOLOv8 network model; step 5: performing model training according to the optical remote sensing image target detection data set and the improved YOLOv8 network model, and determining the optical remote sensing image target detection network model. The remote sensing image target detection method based on an attention mechanism and lightweight convolution of the present invention fuses the attention mechanism and lightweight convolution, has more accurate positioning, reduces redundant information, and has higher detection accuracy.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Apple counting method based on AD-YOLO model and MR-SORT method

    CN118097658A